Google Gemini launchedBack
Today, Google (finally) launched its AI model Gemini, which will compete with ChatGPT. At the same time, the AI chatbot Bard will be upgraded and the chatbot will be rolled out to various Google products in the coming months.
Gemini is based on the technology used for Bard and other services such as the Chrome search engine and web browser. But Gemini is broader: it can handle text, images, audio, video and programming languages.
Three versions of Gemini will be available. The 'Pro' variant will be available to consumers next week in chatbot Bard, for now only in English. This version is said to be better than the free version of ChatGPT (GPT 3.5) that has been available since November last year.
The most extensive variant is called 'Gemini Ultra'. According to Google's own tests, this version performs better in almost all areas than the paid, most extensive version of OpenAI, GPT4. Gemini Ultra is intended to perform very complex tasks. Gemini Pro scales with the difficulty of the task. Finally, there is Gemini Nano, which has limited computing power. This model is so flexible that it can run on all types of hardware 'from data centers to mobile devices'.
Multiple types of data
Gemini can process multiple types of data, and does it very well. Gemini Ultra is the first AI model to outperform human experts in mathematics, physics, history, law, medicine and ethics, among others, in both general knowledge and problem-solving skills.
Although the most powerful version, Ultra, was used for the tests, the possibilities of the other versions of Gemini are also very extensive. All versions are trained from the start to process multiple types of data; something that competitors such as ChatGPT cannot do.
The stakes are very high. CEO Sundar Pichai calls the transition to AI the “most profound in our lifetimes, far bigger than the transition to mobile or the internet.” That is why chatbot Bard will immediately start using Gemini Pro, the version that can reason, plan and understand at a higher level.
Because Google is catching up after it was surprised by the arrival of ChatGPT last year. It fears that it will fall so far behind that it will no longer be relevant in this area. The Gemini version that is now being launched is not even the very best. And it is not without reason that even co-founder Sergey Brin was brought back to work on Gemini. And Google quickly resolved issues with non-English prompts by offering Gemini only in the English language for now.
For the time being, interest in ChatGPT in the Netherlands is much greater than for Bard. And in addition, ChatGPT is also popular as a search engine. Instead of ten links with advertisements, you get a human-looking answer to your search questions. A completely different way of searching, which also raises the question of how appropriate Google's traditional search ads still are. Every step Google takes in this area could jeopardise its own revenue model.
But Google still does not dare to immediately launch the most extensive version on the market. It wants to first test this with a selected group of customers and developers.
Roll out in the coming months
In the coming months, Gemini will be rolled out to services such as Ads and Duet AI, and the AI model will be integrated into the Chrome browser and Google search engine. According to Google, this will produce a faster AI search engine.
Developers can add Gemini 1.0 to their own AI products starting December 13 via the Gemini API in Google AI Studio or Google Cloud Vertex AI. Gemini Ultra will be launched in 2024 for developers and enterprise customers using Google's cloud services.